This project addresses two major bioinformatics problems: the development of better software for finding genes in eukaryotic genome sequences, and the development of genome assemblers for large shotgun sequencing projects. The gene finding project will pursue two tracks: first, we will continue to improve our Generalized Hidden Markov Model and our Pair Hidden Markov Model gene finders, training them for new species as new genomes appear, and enhancing their capabilities to use related species as a guide to gene finding in a new species. Second, we will develop a new eukaryotic annotation pipeline, which will integrate the results from a wide range of sources, including gene finders, protein sequence alignments, cDNA and EST alignments, and other sequence features. This pipeline will be used to predict comprehensive gene sets for multiple species, focusing especially on species for which the available annotation is incomplete or outdated. The pipeline will also be available as a service to annotate genomes for other groups. The assembler project will include several major efforts. First, we will continue to build on our successful open source assembler project, AMOS, adding new modules to allow inter-operation with other assembly packages. Second, we will develop new assemblers that can handle pyrosequencing data, low coverage genome projects, and sequences collected from complex mixtures of species. Third, we will provide assembly services to genome sequencing centers and other collaborators, helping them to assemble genomes using the latest available assembly tools. These will include new sequencing projects as well as genomes that, although already sequenced, can be re-assembled more accurately using improved assembly software. For all of the software development projects, we will continue our practice of making all our source code freely available to investigators in the scientific research community worldwide.